KNOWLEDGE BASE SYSTEM TECHNOLOGY IN 
THE DEFENSE MAPPING AGENCY'S 
DIGITAL PRODUCTION SYSTEM 
Luanne I. Mulawa 
Defense Mapping Agency 
3200 South Second Street 
St Louis, Missouri 63118 
ABSTRACT 
The implementation of Knowledge Base System 
technology for mapping, charting, and geodetic 
production applications is emerging out of its infancy. 
Consequently, literature on Knowledge Base Systems 
similar in size and complexity to the Defense Mapping 
Agency (DMA's) Digital Production System's (DPS's) is 
rare. 
The DMA made a commitment to this applied artificial 
intelligence technology in an effort to secure a 
production capability with sufficient flexibility to respond 
to changing requirements and operational experience. 
The goal of this paper is to provide an overview of the 
DPS with an emphasis on the Data Extraction 
Knowledge Base System. This Knowledge Base 
System was designed to provide decision support for 
automated feature extraction. 
Knowledge engineers, programmers, and project 
managers coordinating, directing, or managing a 
"corporate" KBS development effort should gain 
valuable insight into the potential benefits associated 
with developing a large cartographic KBS for production 
application. 
Key Words: artificial intelligence; digital system; 
knowledge base; feature extraction; mapping 
INTRODUCTION 
The Defense Mapping Agency (DMA) produces and 
distributes mapping, charting, and geodetic products; 
services; and training for the Department of Defense 
(DoD). 
A host of new weapon systems, training systems, and 
intelligence analysis systems are being developed 
which will rely upon digital mapping, charting, and 
geodetic (MC&G) data for operation. The data 
requirements for these new systems are rapidly 
overwhelming current production resources. 
In 1982, DMA initiated an ambitious modernization 
program. There were three fundamental driving forces 
that lead to this modernization program: DMA needed 
an end-to-end digital production system to effectively 
satisfy both existing and emerging requirements for 
digital MC&G data; management needed the 
capabilities to be more flexible and responsive to new 
requirements and technology; and DMA needed to 
increase production throughput and reduce operating 
costs for MC&G production. 
This 10 year development initiative is rapidly 
approaching operational readiness. By the mid 1990s 
DMA mapping and charting will be revolutionized with 
the full deployment of this state-of-the-art cartographic 
production system referred to as the Digital Production 
System (DPS). 
The DPS will provide DMA with a softcopy production 
system that will be integrated into DMA's existing hybrid 
(hardcopy/softcopy) mapping, charting, and geodetic 
production environment. 
DPS's functionality and operations are distributed 
among seven "segments". The Production Management 
Segment (PM/S) controls and monitors DMA production. 
It performs production planning and programming 
activities; generates production plans, assignments, and 
schedules; and manages production resources. The 
Data Services Segment (DS/S) manages system 
communication services; manages temporary and 
archived storage services, and manages the DPS 
MC&G Data Base. 
There are five segments that make up the DPS 
production pipe-line. The first of these segments is the 
Source Acquisition Segment (SA/S). This segment 
provides DPS with the capabilities to manage DMA 
image library holdings and the geodetic control data 
base. The Source Preparation Segment (SP/S) collects 
and evaluates all source materials applicable to a 
specific assignment; conducts point mensuration for 
geopositioning; and assembles assignment packages 
required for DE/S and PG/S production purposes. The 
assignment packages include instructions, 
specifications, and source material listings required to 
complete a production assignment. The Hardcopy 
Exploitation Segment (HE/S) performs triangulation 
(outputs include rectified and ortho-rectified imagery). 
The Data Extraction Segment (DE/S) extracts terrain 
and feature data from softcopy imagery. This data is 
edited and then stored in the MC&G Data Base to 
support product finishing. The last segment in the 
production pipe-line is the Production Generation 
Segment (PG/S). PG/S transforms, integrates, and 
generates graphic and digital data for hardcopy and 
digital product finishing. 
Within DPS there are KBSs that provide decision 
support for planning of production programs; stereo 
extraction of cartographic features; and data 
segregation, symbolization, conflict detection and 
resolution, generalization, and labeling for product 
finishing. These KBSs will play a central role in DPS 
performance and flexibility, and in protecting DPS from 
technological obsolescence. Each KBS was designed 
differently due to segment functional and performance 
requirements, how each KBS operates within the 
segment's overall design, and differing strategies and 
philosophies of segment developers. 
The remainder of this paper will focus on the design, 
operations, testing, and maintenance of DPS's most 
mature large scale production KBS, the Data Extraction 
KBS. This system was developed under contract with 
General Dynamic's Electronic Division. The DE/S KBS 
is a decision aid that will revolutionize feature extraction 
946 
(€ "TS IA “p= ely 
e 
— 
— 
TS OO -* OO 0 frm TS 
^* C) (6 (0 r31— 
==" CB 
SO >77 2050 >90 AH TON oo